David,
After the "git pull", I've rebuilt Poly/ML and ProofPower and had no problems with any combination of --enable-intinf-asint, --enable-compact32bit and MacOS v. Fedora.
Poly/ML compiled with --enable-compact32bit gives me 5-8% improvement in execution times and 30-40% improvement in the size of saved states on some typical ProofPower examples. So that's a worthwhile improvement - many thanks to you and Makarius!
Regards,
Rob.
On 10 Mar 2019, at 16:30, David Matthews <David.Matthews at prolingua.co.uk> wrote:
Rob, Phil and everyone else,
I'm in the process of putting together the 5.8 release and as part of that I've pushed an updated pre-built compiler for X64/32 for Linux, Mac OS etc. That means that for that platform only, for the moment, it is no longer necessary to run "make compiler", although it won't hurt. It should no longer produce the assertion fault, so if it does this is a different bug. To be absolutely sure please run "make distclean" before trying "git pull".
The pre-built compilers (in the "imports" directory) are essentially binary blobs although they are encoded as text. Since they are large and any update results in a major change to the text I try to avoid updating them more often than once for each release.
Regards, David _______________________________________________ polyml mailing list polyml at inf.ed.ac.uk http://lists.inf.ed.ac.uk/mailman/listinfo/polyml